The current parser checks for the "@S" of the "@SP" chiral class and assumes the next character will be a "P".
This means OpenBabel allows the invalid SMILES of "FPo@S92(Br)Cl"
This may also trigger a possible read error/segfault on a string like "[C@S" because it will skip the NUL terminator and try to process the succeeding characters. Worst case is if that happens to be in unallocated memory, which is unlikely.
I regard this as a low-priority task.